PDA

View Full Version : Rybka (operator ggrayggray) v qpawn



Garvinator
06-08-2006, 02:32 PM
1. e4

pax
06-08-2006, 03:29 PM
Interesting. What specifications are you using on Rybka Garvin? Computer specs plus how long you give it per move. Are you calculating while it is the opponent's move?

Garvinator
06-08-2006, 03:41 PM
Interesting. What specifications are you using on Rybka Garvin? Computer specs plus how long you give it per move. Are you calculating while it is the opponent's move?
The choice of engine is Rybka 1.0 Beta 32 Bit. It is the freeware Rybka engine. Also it will be using Nalimov endgame tablebases.

Just using infinite analysis on my move only until I am satisfied that the engine is not going to change its analysis. Anything else?

qpawn
06-08-2006, 04:46 PM
I am happy for Gary to use whatever form of analysis the computer does best. If he uses any deep position analysis this should be used in an objective manner based upon the best numerical outcome that the engine arrives at.

This is a computer vesus human contest; it's not a person versus person plus engine or engine plus person versus engine plus person. I will be using entirely my own analysis.

It is interesting to see how it goes; the galactic chess championship :D

Desmond
06-08-2006, 04:47 PM
Anything else?What are the stakes?

qpawn
06-08-2006, 04:48 PM
Th e only conditions that I have imposed are those that I listed in the other thread.

Bring it on!

qpawn
06-08-2006, 04:50 PM
Stakes? There is no need for such crass debasement of a contest! This bout transcends all financial or material rewards: it is for the pride of the winner to have bested an ultimate foe. Of course I plan on that winner being me :D

qpawn
06-08-2006, 04:51 PM
Rybka Qpawn


1. e4 e6

*****

I am going to blow its silicon brains out French style :D

Garvinator
06-08-2006, 04:55 PM
1. e4 e6 2. d4

qpawn
06-08-2006, 04:57 PM
Rybka versus qpawn:

1. e4 e6
2. d4 d5

Garvinator
06-08-2006, 04:59 PM
1. e4 e6 2. d4 d5 3. Nc3

qpawn
06-08-2006, 06:43 PM
I respond 3...Nf6 making it:

1. e4 e6

2. d4 d5

3. Nc3 Nf6

I haven't got my mind around these pgn things Gary so it would be good if yoy set up the pgn board every few moves.

Garvinator
06-08-2006, 08:04 PM
I respond 3...Nf6 making it:

1. e4 e6

2. d4 d5

3. Nc3 Nf6

I haven't got my mind around these pgn things Gary so it would be good if yoy set up the pgn board every few moves.
The name is Garvin ;)

Please make the moves going across the page instead of down, helps with setting up the pgn board and also saves space.

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5

qpawn
07-08-2006, 07:39 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N ( from f6) to d7

Garvinator
07-08-2006, 08:41 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3

qpawn
08-08-2006, 09:39 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3 c5

Garvinator
08-08-2006, 10:57 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3 c5 6. Nb5

qpawn
09-08-2006, 11:21 AM
Rybka's last move is surely an error ; I had already looked at it. It's a misguided attempt to plonk the knight at d6.

I can't say that Rybka has been very impressive so far :D

I will look over my intended replies again; I have to make absolutely sure that I exploit this error to the max :evil:

pax
09-08-2006, 04:50 PM
Rybka's last move is surely an error ; I had already looked at it. It's a misguided attempt to plonk the knight at d6.

I can't say that Rybka has been very impressive so far :D

I will look over my intended replies again; I have to make absolutely sure that I exploit this error to the max :evil:

Agreed, it's a poor move. But it is almost certainly the fault of the opening book, not of the engine. Garvin: what opening book are you using?

Garvinator
09-08-2006, 05:06 PM
Agreed, it's a poor move. But it is almost certainly the fault of the opening book, not of the engine. Garvin: what opening book are you using?
It isnt the fault of the opening book. This is what the engine is recommending 6. Nb5 using infinite analysis.

For the opening book, I am using the main fritz book with alot of engine games and top gm games added in. I dont like the move Nb5 either, but as I said, I wouldnt input human moves because I think they are better. Just letting the engine decide.

I am letting the engine analyse this position again and see what it comes up with after about 1 hour 'thought'

In the opening book, I have dxc5, Bb5 and Ne2 as leading contenders.I will report back in about an hour with results.

pax
09-08-2006, 05:12 PM
It isnt the fault of the opening book. This is what the engine is recommending 6. Nb5 using infinite analysis.

Well engines without opening books do play notoriously bad openings.

How many moves were played from the opening book? 5 moves seems like a very short horizon.

I guess Rybka has surrendered the first move advantage, but I guess no more than that just yet.

Garvinator
09-08-2006, 05:36 PM
Well engines without opening books do play notoriously bad openings.

How many moves were played from the opening book? 5 moves seems like a very short horizon.

I guess Rybka has surrendered the first move advantage, but I guess no more than that just yet.
I have used infinite analysis since this move. Before that it was opening book stuff.

It still wants Nb5 after 27 minutes.

As an experiment, I just put it in game mode, instead of infinite analysis. The engine immediately played 6. dxc5, so dxc5 would be the opening book move it wants to play.

pax
09-08-2006, 05:50 PM
I have used infinite analysis since this move. Before that it was opening book stuff.

It still wants Nb5 after 27 minutes.

As an experiment, I just put it in game mode, instead of infinite analysis. The engine immediately played 6. dxc5, so dxc5 would be the opening book move it wants to play.

It is unwise to use 'infinite analysis' mode unless you really genuinely have a long time (like overnight). As I mentioned in another thread, if a computer is to make efficient use of it's resources, it really needs to know how much time it's got. In particular, if it thinks it has more time than it really has, then you may get a bad move due to a lopsided search tree.

So if it has an hour, it is better to set it to an hour per move.

Garvinator
09-08-2006, 06:08 PM
It is unwise to use 'infinite analysis' mode unless you really genuinely have a long time (like overnight). As I mentioned in another thread, if a computer is to make efficient use of it's resources, it really needs to know how much time it's got. In particular, if it thinks it has more time than it really has, then you may get a bad move due to a lopsided search tree.

So if it has an hour, it is better to set it to an hour per move.
how do I set the engine to fixed time and get it to work properly ie it will make a move after a set time using fritz 8.

pax
09-08-2006, 06:24 PM
how do I set the engine to fixed time and get it to work properly ie it will make a move after a set time using fritz 8.

Don't know, that depends on Fritz8. I use arena, which is configurable for fixed time per move.

Bill Gletsos
09-08-2006, 06:41 PM
how do I set the engine to fixed time and get it to work properly ie it will make a move after a set time using fritz 8.I believe it is the levels option under the game menu where you can select the fixed time option where you can enter time per move.

Desmond
09-08-2006, 06:59 PM
I'd like to comment, but I'll wait until after the game to avoid any interference.

Garvinator
09-08-2006, 07:41 PM
I believe it is the levels option under the game menu where you can select the fixed time option where you can enter time per move.
yeah I found that earlier, was trying to get it to work properly. Fixed time is not something I use in F8 :(

Anyways, Nb5 played, still awaiting qpawn's response.

qpawn
09-08-2006, 11:21 PM
I have already worked out two adequate "refutations" of Nb5. At best Rybka will maintain equality.

Since this is an experiment I offer to allow Rybka to take the move Nb5 back and play something better. My offer does not have to be accepted. But it is clear that if the engine is using any reasonable computer level/analysis mode etc setting then there is no way that Nb5 would have been played. Even the move before Nf3 was not the best; f4 was better to threaten the thematic advance f5 at some point and to support the centre against a later ...f6. But Nf3 is "passable". Nb5 is a failure. It reminds me of the novag computer I had in the 80s . It would have no idea in closed positions and just move its knights back and forth.

If Garvin does not feel comfortable operating Rybka for this task then maybe someone else would like to step in?

The other issue that I have is that , believe it or not, I have an advantage over the computer. I go very deep in my correspondence analysis [ ask some of my current CCLA opponents! ] . I go beyond teh horizon effect of any engine. For Rybka to have any sort of "correspondence" analysis that's comparable to what I do infinite analysis won't cut it; it will need to do some sort of deep position analysis. Having done some deep position analysis with fritz it usually takes about 6 to 8 hours to do deep analysis. Deep position analysis produces about the same level of depth that I do for correspondence postions.

Some things for Garvin to think about :)

Garvinator
09-08-2006, 11:25 PM
I am still continuing and Nb5 is the move played. Not best but so be it. I am not going to take it back.

How about you make a move in the other game?

qpawn
09-08-2006, 11:27 PM
ok . I am a bit tired now. But first thing tomorrow I will post more moves/responses.

Garvinator
09-08-2006, 11:28 PM
ok . I am a bit tired now. But first thing tomorrow I will post more moves/responses.
you only need to post a move, nothing more.

qpawn
10-08-2006, 12:25 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N[f6]d7 5. Nf3 c5 6. Nb5 a6

Garvinator
10-08-2006, 01:15 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N[f6]d7 5. Nf3 c5 6. Nb5 a6
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N[f6]d7 5. Nf3 c5 6. Nb5 a6 7. Nd6

MichaelBaron
10-08-2006, 01:41 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3


Interesting...5.f4 is more common...I guess Rybka has it "secret" opening book installed.:hmm:

qpawn
11-08-2006, 10:13 AM
Garvin, you moved the wrong knight to d7 in your last pgn.

Anyway, my next move is 7...Bxd6. So:

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N[f6]d7 5. Nf3 c5 6. Nb5 a6 7. Nd6+ Bxd6

Garvinator
11-08-2006, 12:06 PM
Garvin, you moved the wrong knight to d7 in your last pgn.

Anyway, my next move is 7...Bxd6. So:

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 N[f6]d7 5. Nf3 c5 6. Nb5 a6 7. Nd6+ Bxd6
no, you incorrectly notated your knight move for the pgn board and I didnt check it properly. When having the choice of moving either knight to the same square, you just write the letter of the square, not both the square and number. For some reason though, the pgn board moved the b8 knight, which isnt mentioned at all.

Therefore, your 4th move should be written as Nfd7, that is all.

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3 c5 6. Nb5 a6 7. Nd6+ Bxd6 8. exd6

Garvinator
12-08-2006, 04:37 PM
cough

qpawn
12-08-2006, 08:41 PM
Six planes about to be brought down by misguided looney toon terrorists. Football. Dr who. The meaning of life [ and no, it's not 42].

My first priority is not thsi thread; if you want that to be the case please end these two games right now and get annother opponent for Rybka.

Nobody else is prepared to play against a silicon monster. So if I am a bit slow you will have to like it or lump it.

pax
12-08-2006, 09:09 PM
Give the guy a break. Real correspondence chess is often played at 10 moves in 50 days or slower. Settle in for the long haul.

MichaelBaron
14-08-2006, 01:00 AM
It is unwise to use 'infinite analysis' mode unless you really genuinely have a long time (like overnight). As I mentioned in another thread, if a computer is to make efficient use of it's resources, it really needs to know how much time it's got. In particular, if it thinks it has more time than it really has, then you may get a bad move due to a lopsided search tree.

So if it has an hour, it is better to set it to an hour per move.

The purpose of the game is to have a human vs. computer match. Once white starts toying with possible improvements (e.g. adding opening book or "guiding the computer") it will no longer be a human vs. computer competition in its pure form. it will become more like computer+human vs. human :hmm:

pax
14-08-2006, 09:45 AM
The purpose of the game is to have a human vs. computer match. Once white starts toying with possible improvements (e.g. adding opening book or "guiding the computer") it will no longer be a human vs. computer competition in its pure form. it will become more like computer+human vs. human :hmm:

I'm not talking about the human aiding the computer in any way. I'm just saying that the computer parameters need to be set appropriately to the conditions of the match, otherwise the computer will be handicapped.

qpawn
15-08-2006, 01:07 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3 c5 6. Nb5 a6 7. Nd6+ Bxd6 8. exd6 0-0

I don't think that there is any hurry to grab the exiled pawn straight away. But this move required a bit of thought. If white could hook the pawn up with a bishop at e7 , or exchange off a knight and get a supporting pawn, then white may be better. But the solid text move aims to deal with both of those things.

Black has a slight advantage; the loss of the pawn is unavoidable but white can hold it for quite a while. Against a human I might expect a kingside pawn storm to avoid a lost endgame for white; but the computer probably won't think that way.

If white had whacked the knight back to c3 instead of venturing to d6 it would be an equal position in spite of the lost tempo.

Garvinator
15-08-2006, 01:19 PM
1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nf3 c5 6. Nb5 a6 7. Nd6+ Bxd6 8. exd6 0-0 9. Be2

Garvinator
23-08-2006, 11:51 AM
8 days since qpawn's last move, that is a ridiculous amount of time between moves. Even in the currently run tournament, where they are playing 14 games at once, they have agreed to a move rate of 7 days. You have timed out, goodbye.

Desmond
23-08-2006, 02:52 PM
should I make comments now, or will qpawn come back with a move sometime?

Garvinator
23-08-2006, 02:53 PM
should I make comments now, or will qpawn come back with a move sometime?
make your comments.

Desmond
23-08-2006, 03:08 PM
Rybka's last move is surely an error ; I had already looked at it. It's a misguided attempt to plonk the knight at d6.

I can't say that Rybka has been very impressive so far :D

I will look over my intended replies again; I have to make absolutely sure that I exploit this error to the max :evil:qpawn's comment refers to 6. Nb5.

It is not a weak move. It is fact a strong, tactical move. The N cannot be tolerated on d6, which will probably force black to capture Bxd6 sooner or later (as happened in the game). Giving up the bishop is very risky for black, since he will have tremendous difficulty guarding his dark squares.

The pawn (on d6) will be a bone in black's throat. He may be able to win it, but even then it will be a worthwhile investment for white.

If I had the combinative brilliance of Rybka, 6. Nb5 is exactly the move I would have played. I would be very confident of winning in the middlegame, especially in view of qpawn's underestimation of the move.

qpawn
23-08-2006, 06:20 PM
I will be coming back to play the games.

Been a bit busy lately.

But I disagree with Boris's analysis in the strongest possible terms. If the pawn were to be any bone in my throat then rybka would have got in bg5 but I can answer that with ease by...f6. For someone who says that he is a strong player who has played at the high levels of tournaments I am very surprised at such flawed analysis by Boris. If white has achieved anything by putting such a pathetic pawn there then I will have to rewrite the positional understanding of chess that I have.

Garvinator
23-08-2006, 06:31 PM
I will be coming back to play the games.

Been a bit busy lately.

But I disagree with Boris's analysis in the strongest possible terms. If the pawn were to be any bone in my throat then rybka would have got in bg5 but I can answer that with ease by...f6. For someone who says that he is a strong player who has played at the high levels of tournaments I am very surprised at such flawed analysis by Boris. If white has achieved anything by putting such a pathetic pawn there then I will have to rewrite the positional understanding of chess that I have.
well get on with the games instead of whinging. Talk is cheap, I want to see action.

Desmond
23-08-2006, 07:19 PM
But I disagree with Boris's analysis in the strongest possible terms.We shall see...
If the pawn were to be any bone in my throat then rybka would have got in bg5 but I can answer that with ease by...f6. I won't comment on the specifics, as it seems the game is in progress.
For someone who says that he is a strong player who has played at the high levels of tournaments I am very surprised at such flawed analysis by Boris.Strength is relative. I think I am between yourself and Rybka.
If white has achieved anything by putting such a pathetic pawn there then I will have to rewrite the positional understanding of chess that I have.
We shall see...

qpawn
23-08-2006, 09:29 PM
well get on with the games instead of whinging. Talk is cheap, I want to see action.

************

I'm sick of the way you think you can speak to me, that being 1 example.

You have ruined any desire that I had to continue this. My move was going to be 9...c5 x d4 . It opens things up a little for white's 2 bishops but is worth it to simplify things and get the pawn.